[Nut-upsdev] autogen.sh; was Re: Minor issues cross-compiling windows_port branch

Charles Lepple clepple at gmail.com
Wed Jan 26 13:28:20 UTC 2011


On Jan 26, 2011, at 7:32 AM, Paul Fertser wrote:

> On Wed, Jan 26, 2011 at 07:11:59AM -0500, Charles Lepple wrote:
>> On Wed, Jan 26, 2011 at 5:11 AM, Paul Fertser <fercerpav at gmail.com>  
>> wrote:
>>> * missing but required by the configure
>>>  scripts/augeas/nutupsconf.aug.in
>>>  scripts/hal/ups-nut-device.fdi.in
>>>  scripts/udev/nut-usbups.rules.in
>>
>> Currently (r2847), you can use the "nut-autoreconf" script to  
>> generate
>> these files. After the next merge from the trunk, it will be
>> "autogen.sh".
>
> You're right, of course. Somehow i keep forgetting "autogen.sh"-like  
> scripts
> are worth using sometimes. Not sure about the official autotools  
> developers'
> position on that, but to me it feels unclean that one needs an  
> additional
> script instead of autoreconf -i.

If there's a hook that we're overlooking, let us know. (It would be  
nice to incorporate these files into the --enable-maintainer-mode  
Makefile rules, but that kind of recursion makes my head spin.)

In the case of the udev/HAL rules, we are trying to mechanically  
generate the rules from the USB device  information in each driver. I  
think the main reason for the .in files is to substitute the  
@RUN_AS_GROUP@ parameter from ./configure (but I don't know if there  
was another driving force there).

We basically have two classes of people building from source: those  
who build from a tarball (and don't need autoconf, automake, libtool,  
or the Python/Perl libraries needed to regenerate the .in files), and   
those building from source control. If we invert the process to fit  
the autotools framework (by having the configure script create another  
script at configuration time), I think we end up putting additional  
build-time requirements on users.

I haven't had time to look at the Augeas scripts in depth.

Another set of eyes on the code never hurts, though. Let us know if  
we're overlooking an easier way to do this.

-- 
Charles Lepple



More information about the Nut-upsdev mailing list